A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms

Nusrat J. Epsi; John H. Powers; David A. Lindholm; Katrin Mende; Allison Malloy; Anuradha Ganesan; Nikhil Huprikar; Tahaniyat Lalani; Alfred Smith; Rupal M. Mody; Milissa U. Jones; Samantha E. Bazan; Rhonda E. Colombo; Christopher J. Colombo; Evan C. Ewers; Derek T. Larson; Catherine M. Berjohn; Carlos J. Maldonado; Paul W. Blair; Josh Chenoweth; David L. Saunders; Jeffrey Livezey; Ryan C. Maves; Margaret Sanchez Edwards; Julia S. Rozman; Mark P. Simons; David R. Tribble; Brian K. Agan; Timothy H. Burgess; Simon D. Pollett; Jessica J. Cowden; Teresa M. Merritt; Nora Elnahas; Christa Glinn; Donna Jennings; Celia Byrne; Jennifer Rusiecki; Ann Scher; D. Lindholm; C. Colombo; R. Colombo; C. Mount; C. Schofield; M. Stein; T. Lalani; C. Berjohn; K. Chung; C. Olsen; S. Richard; M. Simons

doi:10.1371/journal.pone.0281272

A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms

Nusrat J. Epsi, John H. Powers, David A. Lindholm, Katrin Mende, Allison Malloy, Anuradha Ganesan, Nikhil Huprikar, Tahaniyat Lalani, Alfred Smith, Rupal M. Mody, Milissa U. Jones, Samantha E. Bazan, Rhonda E. Colombo, Christopher J. Colombo, Evan C. Ewers, Derek T. Larson, Catherine M. Berjohn, Carlos J. Maldonado, Paul W. Blair, Josh ChenowethDavid L. Saunders, Jeffrey Livezey, Ryan C. Maves, Margaret Sanchez Edwards, Julia S. Rozman, Mark P. Simons, David R. Tribble, Brian K. Agan, Timothy H. Burgess, Simon D. Pollett, Jessica J. Cowden, Teresa M. Merritt, Nora Elnahas, Christa Glinn, Donna Jennings, Celia Byrne, Jennifer Rusiecki, Ann Scher, D. Lindholm, C. Colombo, R. Colombo, C. Mount, C. Schofield, M. Stein, T. Lalani, C. Berjohn, K. Chung, C. Olsen, S. Richard^*, M. Simons

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

9 Scopus citations

Abstract

BACKGROUND: Accurate COVID-19 prognosis is a critical aspect of acute and long-term clinical management. We identified discrete clusters of early stage-symptoms which may delineate groups with distinct disease severity phenotypes, including risk of developing long-term symptoms and associated inflammatory profiles.

METHODS: 1,273 SARS-CoV-2 positive U.S. Military Health System beneficiaries with quantitative symptom scores (FLU-PRO Plus) were included in this analysis. We employed machine-learning approaches to identify symptom clusters and compared risk of hospitalization, long-term symptoms, as well as peak CRP and IL-6 concentrations.

RESULTS: We identified three distinct clusters of participants based on their FLU-PRO Plus symptoms: cluster 1 ("Nasal cluster") is highly correlated with reporting runny/stuffy nose and sneezing, cluster 2 ("Sensory cluster") is highly correlated with loss of smell or taste, and cluster 3 ("Respiratory/Systemic cluster") is highly correlated with the respiratory (cough, trouble breathing, among others) and systemic (body aches, chills, among others) domain symptoms. Participants in the Respiratory/Systemic cluster were twice as likely as those in the Nasal cluster to have been hospitalized, and 1.5 times as likely to report that they had not returned-to-activities, which remained significant after controlling for confounding covariates (P < 0.01). Respiratory/Systemic and Sensory clusters were more likely to have symptoms at six-months post-symptom-onset (P = 0.03). We observed higher peak CRP and IL-6 in the Respiratory/Systemic cluster (P < 0.01).

CONCLUSIONS: We identified early symptom profiles potentially associated with hospitalization, return-to-activities, long-term symptoms, and inflammatory profiles. These findings may assist in patient prognosis, including prediction of long COVID risk.

Original language	English
Article number	e0281272
Pages (from-to)	e0281272
Journal	PLoS ONE
Volume	18
Issue number	2 February
DOIs	https://doi.org/10.1371/journal.pone.0281272 https://doi.org/10.1371/journal.pone.0281272
State	Published - Feb 2023
Externally published	Yes

Keywords

Humans
COVID-19/epidemiology
SARS-CoV-2
Post-Acute COVID-19 Syndrome
Interleukin-6
Phenotype
Hospitalization
Machine Learning

Access to Document

Cite this

Epsi, N. J., Powers, J. H., Lindholm, D. A., Mende, K., Malloy, A., Ganesan, A., Huprikar, N., Lalani, T., Smith, A., Mody, R. M., Jones, M. U., Bazan, S. E., Colombo, R. E., Colombo, C. J., Ewers, E. C., Larson, D. T., Berjohn, C. M., Maldonado, C. J., Blair, P. W., ... Simons, M. (2023). A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms. PLoS ONE, 18(2 February), e0281272. Article e0281272. https://doi.org/10.1371/journal.pone.0281272, https://doi.org/10.1371/journal.pone.0281272

@article{b3fd5c530a404e84a6514226d3052dc3,

title = "A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms",

abstract = "BACKGROUND: Accurate COVID-19 prognosis is a critical aspect of acute and long-term clinical management. We identified discrete clusters of early stage-symptoms which may delineate groups with distinct disease severity phenotypes, including risk of developing long-term symptoms and associated inflammatory profiles.METHODS: 1,273 SARS-CoV-2 positive U.S. Military Health System beneficiaries with quantitative symptom scores (FLU-PRO Plus) were included in this analysis. We employed machine-learning approaches to identify symptom clusters and compared risk of hospitalization, long-term symptoms, as well as peak CRP and IL-6 concentrations.RESULTS: We identified three distinct clusters of participants based on their FLU-PRO Plus symptoms: cluster 1 ({"}Nasal cluster{"}) is highly correlated with reporting runny/stuffy nose and sneezing, cluster 2 ({"}Sensory cluster{"}) is highly correlated with loss of smell or taste, and cluster 3 ({"}Respiratory/Systemic cluster{"}) is highly correlated with the respiratory (cough, trouble breathing, among others) and systemic (body aches, chills, among others) domain symptoms. Participants in the Respiratory/Systemic cluster were twice as likely as those in the Nasal cluster to have been hospitalized, and 1.5 times as likely to report that they had not returned-to-activities, which remained significant after controlling for confounding covariates (P < 0.01). Respiratory/Systemic and Sensory clusters were more likely to have symptoms at six-months post-symptom-onset (P = 0.03). We observed higher peak CRP and IL-6 in the Respiratory/Systemic cluster (P < 0.01).CONCLUSIONS: We identified early symptom profiles potentially associated with hospitalization, return-to-activities, long-term symptoms, and inflammatory profiles. These findings may assist in patient prognosis, including prediction of long COVID risk.",

keywords = "Humans, COVID-19/epidemiology, SARS-CoV-2, Post-Acute COVID-19 Syndrome, Interleukin-6, Phenotype, Hospitalization, Machine Learning",

author = "Epsi, {Nusrat J.} and Powers, {John H.} and Lindholm, {David A.} and Katrin Mende and Allison Malloy and Anuradha Ganesan and Nikhil Huprikar and Tahaniyat Lalani and Alfred Smith and Mody, {Rupal M.} and Jones, {Milissa U.} and Bazan, {Samantha E.} and Colombo, {Rhonda E.} and Colombo, {Christopher J.} and Ewers, {Evan C.} and Larson, {Derek T.} and Berjohn, {Catherine M.} and Maldonado, {Carlos J.} and Blair, {Paul W.} and Josh Chenoweth and Saunders, {David L.} and Jeffrey Livezey and Maves, {Ryan C.} and Edwards, {Margaret Sanchez} and Rozman, {Julia S.} and Simons, {Mark P.} and Tribble, {David R.} and Agan, {Brian K.} and Burgess, {Timothy H.} and Pollett, {Simon D.} and Cowden, {Jessica J.} and Merritt, {Teresa M.} and Nora Elnahas and Christa Glinn and Donna Jennings and Celia Byrne and Jennifer Rusiecki and Ann Scher and D. Lindholm and C. Colombo and R. Colombo and C. Mount and C. Schofield and M. Stein and T. Lalani and C. Berjohn and K. Chung and C. Olsen and S. Richard and M. Simons",

note = "Funding Information: This work was supported by awards from the Defense Health Program (HU00012020067) and the National Institute of Allergy and Infectious Disease (HU00011920111). The protocol was executed by the Infectious Disease Clinical Research Program (IDCRP), a Department of Defense (DoD) program executed by the Uniformed Services University of the Health Sciences (USUHS) through a cooperative agreement by the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc. (HJF). This project has been funded in part by the National Institute of Allergy and Infectious Diseases at the National Institutes of Health, under an interagency agreement (Y1-AI-5072). Publisher Copyright: {\textcopyright} 2023 Public Library of Science. All rights reserved.",

year = "2023",

month = feb,

doi = "10.1371/journal.pone.0281272",

language = "English",

volume = "18",

pages = "e0281272",

journal = "PLoS ONE",

issn = "1932-6203",

number = "2 February",

}

Epsi, NJ, Powers, JH, Lindholm, DA, Mende, K, Malloy, A, Ganesan, A, Huprikar, N, Lalani, T, Smith, A, Mody, RM, Jones, MU, Bazan, SE, Colombo, RE, Colombo, CJ, Ewers, EC, Larson, DT, Berjohn, CM, Maldonado, CJ, Blair, PW, Chenoweth, J, Saunders, DL, Livezey, J, Maves, RC, Edwards, MS, Rozman, JS, Simons, MP, Tribble, DR , Agan, BK , Burgess, TH , Pollett, SD, Cowden, JJ, Merritt, TM, Elnahas, N, Glinn, C, Jennings, D, Byrne, C , Rusiecki, J, Scher, A, Lindholm, D , Colombo, C , Colombo, R, Mount, C, Schofield, C , Stein, M , Lalani, T , Berjohn, C, Chung, K, Olsen, C , Richard, S & Simons, M 2023, 'A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms', PLoS ONE, vol. 18, no. 2 February, e0281272, pp. e0281272. https://doi.org/10.1371/journal.pone.0281272, https://doi.org/10.1371/journal.pone.0281272

A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms. / Epsi, Nusrat J.; Powers, John H.; Lindholm, David A. et al.
In: PLoS ONE, Vol. 18, No. 2 February, e0281272, 02.2023, p. e0281272.

Research output: Contribution to journal › Article › peer-review

TY - JOUR

T1 - A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms

AU - Epsi, Nusrat J.

AU - Powers, John H.

AU - Lindholm, David A.

AU - Mende, Katrin

AU - Malloy, Allison

AU - Ganesan, Anuradha

AU - Huprikar, Nikhil

AU - Lalani, Tahaniyat

AU - Smith, Alfred

AU - Mody, Rupal M.

AU - Jones, Milissa U.

AU - Bazan, Samantha E.

AU - Colombo, Rhonda E.

AU - Colombo, Christopher J.

AU - Ewers, Evan C.

AU - Larson, Derek T.

AU - Berjohn, Catherine M.

AU - Maldonado, Carlos J.

AU - Blair, Paul W.

AU - Chenoweth, Josh

AU - Saunders, David L.

AU - Livezey, Jeffrey

AU - Maves, Ryan C.

AU - Edwards, Margaret Sanchez

AU - Rozman, Julia S.

AU - Simons, Mark P.

AU - Tribble, David R.

AU - Agan, Brian K.

AU - Burgess, Timothy H.

AU - Pollett, Simon D.

AU - Cowden, Jessica J.

AU - Merritt, Teresa M.

AU - Elnahas, Nora

AU - Glinn, Christa

AU - Jennings, Donna

AU - Byrne, Celia

AU - Rusiecki, Jennifer

AU - Scher, Ann

AU - Lindholm, D.

AU - Colombo, C.

AU - Colombo, R.

AU - Mount, C.

AU - Schofield, C.

AU - Stein, M.

AU - Lalani, T.

AU - Berjohn, C.

AU - Chung, K.

AU - Olsen, C.

AU - Richard, S.

AU - Simons, M.

N1 - Funding Information: This work was supported by awards from the Defense Health Program (HU00012020067) and the National Institute of Allergy and Infectious Disease (HU00011920111). The protocol was executed by the Infectious Disease Clinical Research Program (IDCRP), a Department of Defense (DoD) program executed by the Uniformed Services University of the Health Sciences (USUHS) through a cooperative agreement by the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc. (HJF). This project has been funded in part by the National Institute of Allergy and Infectious Diseases at the National Institutes of Health, under an interagency agreement (Y1-AI-5072). Publisher Copyright: © 2023 Public Library of Science. All rights reserved.

PY - 2023/2

Y1 - 2023/2

N2 - BACKGROUND: Accurate COVID-19 prognosis is a critical aspect of acute and long-term clinical management. We identified discrete clusters of early stage-symptoms which may delineate groups with distinct disease severity phenotypes, including risk of developing long-term symptoms and associated inflammatory profiles.METHODS: 1,273 SARS-CoV-2 positive U.S. Military Health System beneficiaries with quantitative symptom scores (FLU-PRO Plus) were included in this analysis. We employed machine-learning approaches to identify symptom clusters and compared risk of hospitalization, long-term symptoms, as well as peak CRP and IL-6 concentrations.RESULTS: We identified three distinct clusters of participants based on their FLU-PRO Plus symptoms: cluster 1 ("Nasal cluster") is highly correlated with reporting runny/stuffy nose and sneezing, cluster 2 ("Sensory cluster") is highly correlated with loss of smell or taste, and cluster 3 ("Respiratory/Systemic cluster") is highly correlated with the respiratory (cough, trouble breathing, among others) and systemic (body aches, chills, among others) domain symptoms. Participants in the Respiratory/Systemic cluster were twice as likely as those in the Nasal cluster to have been hospitalized, and 1.5 times as likely to report that they had not returned-to-activities, which remained significant after controlling for confounding covariates (P < 0.01). Respiratory/Systemic and Sensory clusters were more likely to have symptoms at six-months post-symptom-onset (P = 0.03). We observed higher peak CRP and IL-6 in the Respiratory/Systemic cluster (P < 0.01).CONCLUSIONS: We identified early symptom profiles potentially associated with hospitalization, return-to-activities, long-term symptoms, and inflammatory profiles. These findings may assist in patient prognosis, including prediction of long COVID risk.

AB - BACKGROUND: Accurate COVID-19 prognosis is a critical aspect of acute and long-term clinical management. We identified discrete clusters of early stage-symptoms which may delineate groups with distinct disease severity phenotypes, including risk of developing long-term symptoms and associated inflammatory profiles.METHODS: 1,273 SARS-CoV-2 positive U.S. Military Health System beneficiaries with quantitative symptom scores (FLU-PRO Plus) were included in this analysis. We employed machine-learning approaches to identify symptom clusters and compared risk of hospitalization, long-term symptoms, as well as peak CRP and IL-6 concentrations.RESULTS: We identified three distinct clusters of participants based on their FLU-PRO Plus symptoms: cluster 1 ("Nasal cluster") is highly correlated with reporting runny/stuffy nose and sneezing, cluster 2 ("Sensory cluster") is highly correlated with loss of smell or taste, and cluster 3 ("Respiratory/Systemic cluster") is highly correlated with the respiratory (cough, trouble breathing, among others) and systemic (body aches, chills, among others) domain symptoms. Participants in the Respiratory/Systemic cluster were twice as likely as those in the Nasal cluster to have been hospitalized, and 1.5 times as likely to report that they had not returned-to-activities, which remained significant after controlling for confounding covariates (P < 0.01). Respiratory/Systemic and Sensory clusters were more likely to have symptoms at six-months post-symptom-onset (P = 0.03). We observed higher peak CRP and IL-6 in the Respiratory/Systemic cluster (P < 0.01).CONCLUSIONS: We identified early symptom profiles potentially associated with hospitalization, return-to-activities, long-term symptoms, and inflammatory profiles. These findings may assist in patient prognosis, including prediction of long COVID risk.

KW - Humans

KW - COVID-19/epidemiology

KW - SARS-CoV-2

KW - Post-Acute COVID-19 Syndrome

KW - Interleukin-6

KW - Phenotype

KW - Hospitalization

KW - Machine Learning

UR - http://www.scopus.com/inward/record.url?scp=85147834369&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0281272

DO - 10.1371/journal.pone.0281272

M3 - Article

C2 - 36757946

AN - SCOPUS:85147834369

SN - 1932-6203

VL - 18

SP - e0281272

JO - PLoS ONE

JF - PLoS ONE

IS - 2 February

M1 - e0281272

ER -

Epsi NJ, Powers JH, Lindholm DA, Mende K, Malloy A, Ganesan A et al. A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms. PLoS ONE. 2023 Feb;18(2 February):e0281272. e0281272. doi: 10.1371/journal.pone.0281272, 10.1371/journal.pone.0281272

A machine learning approach identifies distinct early-symptom cluster phenotypes which correlate with hospitalization, failure to return to activities, and prolonged COVID-19 symptoms

Abstract

Keywords

Access to Document

Fingerprint

Cite this